4
\$\begingroup\$

An SMPS power supply unit namely Vsupply output can be composed of its desired DC output Vdc, differential mode noise Vdm and common-mode noise Vcm.

If we have such a power supply and an earth grounded oscilloscope(with passive probe), is there a practical way to check whether there exists common-mode noise?

I draw the following diagram for ease to depict the power supply, scope and the nodes. Can there be a way to manifest the existence of common mode noise by using a resistor in between some of the nodes or any other way? I have never tried to quantify how much part of the noise is DM and how much is CM and have no practical experience on it.

Edit for an answer:

1) Where is the probe tip connected?

enter image description here

2) Can we measure CM by this way?:

enter image description here

enter image description here

Edit 2 for a comment:

enter image description here

enter image description here

LAST EDIT:

Scope screen shots: (Ch1 and Ch2 are AC coupled)

Pink is the function waveform:

Ch1-Ch2:

enter image description here

Ch1+Ch2

enter image description here

\$\endgroup\$
  • 2
    \$\begingroup\$ I feel very strongly that you are on the wrong track here. Can you explain what is your overall goal? You can measure voltage between any two points using an oscilloscope. In some cases you may need special probes such as differential probes. But I don't think a power supply generally can be modeled as a combination of common mode and differential mode voltages. I think that is a very artificial model, and the node labelled 'A' in your first diagram doesn't really exist. \$\endgroup\$ – mkeith Oct 4 '19 at 0:14
  • \$\begingroup\$ My goal is to see only CM noise and only DM noise on scope. I added a 1mOhm resistor between A and D. See my last edit. \$\endgroup\$ – cm64 Oct 4 '19 at 0:18
  • 1
    \$\begingroup\$ In your edit you have constructed something in a simulation environment that does not correspond to a real circuit. Sorry, but this does not make any sense. If you want to look at common mode in a real world environment, such as common mode voltage or current flowing in a cable, that is possible. You can use a common mode RF current probe and an oscilloscope or spectrum analyzer. But I am afraid there is some big misunderstanding on your part. You shouldn't be asking what you are asking. I wish you could take a step back and talk about what your real goals are and what you are trying to do. \$\endgroup\$ – mkeith Oct 4 '19 at 5:38
1
\$\begingroup\$

In an SMPS on the DC power supply side, Common Mode (CM) noise occurs when current exits the circuit and returns through the ground. Typically a DC supply is not grounded in two places (although it could be) as shown in the circuit above, this creates a ground loop and causes problems.

Typically noise exits a device through capacitance and then returns through the ground, which is entirely possible with an SMPS because it also produces AC noise that can readily radiate to nearby conductors from the circuit (Vn can also be located in the supply as not shown below, but has the same result):

enter image description here

Source: https://micro.rohm.com/en/techweb/knowledge/emc/s-emc/01-s-emc/6899

As shown in the diagram above, if it truly is the way you have set things up with a ground loop, there would be a few ways to measure the CM noise. One would be to put a scope ground below A (on A's ground), and the other on D. The differential could be measured from point A to D.

enter image description here

Or if you had a good current probe you could use it to probe the CM anywhere around the other loop (if there are not to grounds, you could probe through A's ground).

enter image description here

One thing to keep in mind is Vcm will change with the load, because the switching period and current changes, the noise will also change. So if you are testing for noise, you may want to test over a range of load conditions.

Maybe Regulatory?

If this is for regulatory testing, then typically AC conducted emissions are tested with an SMPS (typically EN55011, non-paywalled description here). The test setup usually involves an LISN and then measuring currents coming to and from the LISN.

enter image description here enter image description here Source: https://www.researchgate.net/publication/224130690_Systematic_Electromagnetic_Interference_Filter_Design_Based_on_Information_From_In-Circuit_Impedance_Measurements

Edit:

If you want to measure the common mode and differential mode without a load, then this is how it could be done with a two channel scope:

enter image description here

The scope ground needs to be connected to the chassis ground of the supply, this is where the noise currents return to the supply. To measure common mode noise, the supply needs a chassis ground.

The differential mode is:

\$V_{DM}=Ch1-Ch2\$

The common mode is:

\$V_{CM}=\frac{Ch1+Ch2}{2}\$

If you didn't have two probes then you could measure and subtract on paper. If you had two probes then you could see both simultaneously which would be better.

Better than that: many digital scopes have averaging and subtraction functions built in. If so then look at the special functions of your scope. You could average Ch1 and Ch2 together and get the Common Mode voltage. Subtracting both would give you the Differential Mode voltage.

My simulation works just fine, the only caveat is the pk-pk voltages don't average and you get 3V instead of 2.5V because one of the AC signals is riding on top of the other one for the Vcm measurement.

Vcm = 1 Vdm = 4

Vcm measured = 3 (should be 2.5) Vdm measured = 4

enter image description here

\$\endgroup\$
  • \$\begingroup\$ Thanks but something is implicit. In my question node C is call it scope's Ch1's positive input(inner pin of the BNC) and node D is the Ch1 ground(scope analog ground) and 1Meg depicts the scope's input impedance. So please see my edit. I have two questions here. 1) Can you please show where will the probe tip and probe ground be connected in your DM mode suggestion? i.stack.imgur.com/bPIuL.jpg and 2) I was thinking a way of neasirn Vcm this way what do you think? i.stack.imgur.com/apZSR.jpg \$\endgroup\$ – cm64 Oct 3 '19 at 21:15
  • \$\begingroup\$ Actually, I did a measurement with USB a while back that gave me an idea. I edited the question. If you have a scope with averaging and subtraction functions, then you should be good to go. \$\endgroup\$ – Voltage Spike Oct 3 '19 at 22:02
  • \$\begingroup\$ If you reread my question (the bounty section), I mentioned already: "... And Ch1-C2 function doesn’t display anything look like DM noise. " I tried to subtract or add I get very weird results. It only works in theory. \$\endgroup\$ – cm64 Oct 3 '19 at 22:08
  • \$\begingroup\$ But I will retry again tomorrow, maybe I didnt connect the probe ground to supply. But where does the scope ground connect here?: i.stack.imgur.com/s2zBU.png Is thet the earth of the supply ac input or DC negative terminal? \$\endgroup\$ – cm64 Oct 3 '19 at 22:12
  • \$\begingroup\$ What do you mean by very weird results? I would expect for the supply to have say CH1 near the supply (lets say 12V) and CH2 near ground, like 10mV. This would lead the DM voltage to be 11.99V and the Common mode voltage to be 6.005V (At DC, the AC measurements will be different) \$\endgroup\$ – Voltage Spike Oct 3 '19 at 22:12
0
\$\begingroup\$

When you measure by ordinary 10:1 probe method you get Vdm and Vcm and stray noise pickup on coax and ground clip from LC resonance from inputs < 20ns rise time. Thus resulting in poor signal capture test method.

1st try to measure the Vdm (Diff mode) noise accurately, then when you capture Vdm +Vcm with a single 10:1 to your reference earth ground you can see the Vcm noise.

Suggested methods for Vdm:

-1 Using two passive 10:1 probes balanced to flat line in A-B (with 20MHz BW on)
-2 Using 200 MHz Fet Buffered Diff Probes - expensive but accurate
-3 Using above but with tip & clip replaced with 5mm spring probes ( easiest solution)
-4 Add small ceramic cap and 50 ( or 49.9) R to test point or direct 1:1 coax connection to DSO using internal 50R term (opt.) or external BNC-T and 50 Ohm BNC terminator. (adapt source with suitable connections:BNC,SMA,solder wire pins (logic diagram shown below)

schematic

simulate this circuit – Schematic created using CircuitLab

Then test using active or dummy loads for load test 10~90%, 10~50%, 50~0% etc for Vdm dynamic load error. (use DSO 20 MHz filter to analyze if fuzzy)

To investigate Vcm thru 2 wire AC-DC converter leakage and Vcm introduced say by 3 prong charger and transformer C pf Leakage through laptop USB cable now this is external Vcm noise. This can also be 3 pronged EMI Line filter leakage current to Earth Gnd.

To investigate Vcm noise thru 2 wire AC-DC converter, you may be observing Vcm noise from any other source connected to load such as leakage from a Laptop USB cable with AC-charger on or a PC Tower with a USB cable to DUT with PSU Vcm issues. Vcm introduced say by 3 prong charger will have a CLC PI EMI line filter leakage to Earth Gnd.

Use Twisted pairs for all noise transient egressors (inductive) and high impedance ingress points, and coil cables to see if there is cable ingress and orient at 90 deg.

Additional Vcm reduction can be done with CLC baluns selected for the spectrum of noise observed. Since these span <3 decades of f, choosing the right Baluns makes a difference on DC cables and high impedance signals to ADC or DSO. Having 120 dB CMRR in an INA is wasted if your interface wires are balanced to 1% with 40 dB CMRR. Ref samples: https://www.schaffner.com/products/emcemi/

Note that when we use Earth grounds, they are only considered 0V at line frequency and are inductive for RF, so depending on layout , never assume all earth grounds are the same in your RF spectrum.

\$\endgroup\$
  • \$\begingroup\$ those who downvote without comments are voting this ignorance. Vcm can originate from both internal and external noise \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Oct 7 '19 at 9:00
-2
\$\begingroup\$

Based on your schematic, if you want to measure \$V_{cm}\$, there is no reason to connect anything except for a load to terminal B of the supply at all, just measure terminal A relative to the ground terminal of the supply.

Even single-ended passive probes usually have both a signal and a ground connection. You would connect the probe ground with as short a wire as possible to the supply ground.

Measurement of \$V_{DM}\$:

schematic

simulate this circuit – Schematic created using CircuitLab

Measurement of \$V_{CM}\$:

schematic

simulate this circuit

\$\endgroup\$
  • \$\begingroup\$ A is the negative terminal; B is the positive terminal. On the left side the ground symbol is earth as in the right side which is earth of the scope. \$\endgroup\$ – cm64 Oct 3 '19 at 18:24
  • \$\begingroup\$ Can you add some diagrams showing the coupling between the supply and the scope to measure DM and CM noise? Thats the question. \$\endgroup\$ – cm64 Oct 3 '19 at 18:28
  • \$\begingroup\$ uk.rs-online.com/web/p/din-rail-power-supplies/0428433 and tracopower.com/products/tmp.pdf couple of examples. Im looking for a general method \$\endgroup\$ – cm64 Oct 3 '19 at 18:34
  • \$\begingroup\$ @cm64: I've added some diagrams. I'll admit it's a very simple method. Have you tried it and found it to be not good enough resolution? \$\endgroup\$ – Justin Oct 3 '19 at 18:43
  • \$\begingroup\$ Im not sure about your first method is correct. The CM noise will flow through the probe ground and the earth and will appear as noise The probe ground an the earth path has resistance it will develop voltage. Im not sure about your second one as well. But first one doesnt look like a diff mode measurement to me. I want to see only DM and then only CM mode on the scope screen. \$\endgroup\$ – cm64 Oct 3 '19 at 18:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.